仅从“ git diff”获取静态代码分析器消息？

Question

This question was migrated from Server Fault because it can be answered on Stack Overflow. Migrated 4 years ago .

In our build system (with sources under git version control), I'd like to get the static code analyzer ( pylint , in this case) messages for every build. And I want them incremental: in the new build report, only the messages introduced by new commits are shown.

I can easily get the 'old' and 'new' commits. Then, the general path is to run the analyzer on the 'new' commit, and then, for each source code line with a message, find whether that line is new/modified, or if it is intact from the 'old' commit. Then, print only messages from new/modified lines.

Is there a simpler way of achieving this goal for pylint ? Or is there an existing implementation of something similar?

Answer 1

You can get list of files changed between commits and pass it to pylint then. With github pullrequest workflow it looks like this:

git --no-pager diff --name-only FETCH_HEAD $(git merge-base FETCH_HEAD master) | grep .py | xargs pylint

Answer 2

The git diff-tree command has a --diff-filter argument that can be used to filter for a specific type of modification:

From the git diff-tree documentation :

--diff-filter=[(A|C|D|M|R|T|U|X|B)…​[*]]
Select only files that are Added (A), Copied (C), Deleted (D), Modified (M), Renamed (R), have their type (ie regular file, symlink, submodule, …​) changed (T), are Unmerged (U), are Unknown (X), or have had their pairing Broken (B). Any combination of the filter characters (including none) can be used.

If you want to get a list of the names of all added files you could use a command like this:

git diff-tree -r --name-only --diff-filter A <oldCommit>..<newCommit>

The -r option in this case means to diff the trees recursively. The --name-only option causes only the file names to be printed, without their status and prevents printing of the diff header.

Since this is one of the git diff commands, it also takes additional arguments that allow filtering the file list. If you are only interested in files ending in '.py' at a certain path you could add -- 'path/*.py' to the command.

If the list of files is sent further to a tool like xargs , it may make sense to add the -z option. This option causes the list of filenames to be delimited by a null byte instead of a newline. This makes the command robust against various special characters that may occur in file names (including spaces). The receiving end then needs to made aware of that. For xargs the option to use would be --null .

A complete command that runs pylint on added files with the '.py' extension between two commits could therefore look like this:

git diff-tree -r -z --name-only --diff-filter A <oldCommit>..<newCommit> -- '*.py' | xargs --null pylint